1. Field of the Invention
The present invention relates to a method of creating a simulation model and the like.
2. Description of the Related Art
According to advance of a semiconductor integrated circuit to high integration, lithography simulation becomes important increasingly. In order to perform lithography simulation with high precision, it is important to use a simulation model with a high precision.
For creating a lithography simulation model, it is necessary to optimize a parameter for the lithography simulation model so as to reduce a size difference between a line width (experimental value) of a photoresist pattern formed actually and a line width (simulation value) obtained by a simulation. As the method for optimizing a parameter for a lithography simulation model, methods using a regression analysis has been known (for example, see “SPIE vol. 2726, p. 209-221, Mar. 13, 1996”).
As one of the methods using a regression analysis, there is a method for determining a function representing a correlation between a parameter of an optical image and an empirical threshold according to a regression analysis. The parameter of an optical image influences a shape or a size of a photoresist pattern. For example, a slope of an intensity distribution of an optical image can be used as the parameter for an optical image. The empirical threshold is an intensity value defined according to a CD value on an intensity distribution of an optical image.
However, when a number of various test patterns are used for improving precision of a simulation model, such a problem that a clear correlation cannot be found between a parameter of an optical image and an empirical threshold, such a problem that the correlation between a parameter of an optical image and an empirical threshold lowers occurs. In such a case, precision of the lithography simulation will deteriorate.
Thus, when a lithography simulation model is created using the regression analysis, there is such a problem that a clear correlation cannot be found between a parameter of an optical image and an empirical threshold, or such a problem that the correlation between a parameter of an optical image and an empirical threshold lowers. Therefore, it was conventionally difficult to obtain a simulation model with a high precision.